Research For Online Fault Recognition And Diagnosis Method Of Distribution Line

Power line is very important in power system, and it takes the task of electricpower transmission. At the same time, power line is quite fragile, any damage fromoutside forces and wrong operation from staff, may cause widespread blackouts, andeven power grid collapse, which will bring great harm to the local economicdevelopment, people’s life and social stability. Therefore, rapid detection and faultdiagnosis of power line fault has important significance for treating failure, restoring thenormal operation of power system, maintaining the security of power system andeconomic interests of users.Fault detection and diagnosis for transmission line and distribution line is alwaysan interesting topic for power engineering personnel. More researches are placed ontransmission lines at home and abroad, and neglected distribution line. At present, thestudy of distribution line emphasis on diagnosis and location of line failure, whichbelong to offline fault diagnosis. Existing monitoring devices and positioning devicescan not monitor real-time situation.of distribution line, can not detect transient faultsand high impedance faults, and can not predict early failure of distribution line. If earlyfaiure can not be ruled out, it will develop into short-circuit fault, which will bring harmto safe operation of power system.Online monitoring, diagnosis and fault prediction system of distribution lines is notrun in real system, in this case, this paper proposed a set of system solution includingonline state monitoring, diagnosis and foresting using the steady-state characteristicsand transient characteristics of distribution line fault information. A monitoring device,which is installed in the main line branch, is developed together with company. Themain contents of research include: The fault feature vectors are extracted byHilbert-Huang transform from voltage and current information; In order to determinefault type, HPL method is proposed to carry on preliminary classification and detailedclassification; Finally, zero-sequence instantaneous power direction is used to identifythe fault line, the results of fault diagnosis are transmitted by a wireless to powerdispatch staff, this results provide the basis for decision making. Details and mainresults are as follows:(1) According to two facts that distribution line fault signal contains thenon-fundamental frequency transient information and Hilbert-Huang transform is applicable to analysize nonlinear, non-stationary signals, this paper proposed a systemfor online monitoring, classification, recognition of distribution line.(2) The paper first proposed the use of Hilbert-Huang Transform method toextract transient information of the distribution line fault signal, that is, decompose thezero sequence voltage signal into intrinsic mode functions, obtain instantaneousamplitude of the first order intrinsic mode fuction using Hilbert transform, calculate thesummation of sampling value in one cycle, which is considered as criterion to judge linefault in distribution system. The results showed that: this method is not only able todetect high impedance fault and intermittent fault, which is difficult to effectivelyextract the fault characteristics of distribution line, it is effective to single phase toground, the two-phase short-circuit grounding fault and so on.(3) This paper first proposed HPL method, which is used to premiliaryclassification and detailed classification for distribution line fault type. Three-phasevoltage and three-phase current are obtained from monitoring device. Once fault occurs,the high frequency components of fault signal are extracted from three-phase voltageand three-phase current using Hilbert-Huang transform, respectively. The instantaneousamplitude of the first two-order intrinsic mode fuction are used to form fault featurevectors, the dimensionality reduction of fault feature vectors is implemented byprincipal component analysis approach, the principal component obtained are used asthe input vector to least squares support vector machines for various fault featuresidentification, determine the fault type of the distribution line. The results show that:this method can not only recognize high impedance faults, intermittent faults and lowresistance faults, but also recognize single-phase short-circuit fault, two-phaseshort-circuit, two-phase short circuit to ground, three-phase short-circuit or three-phaseshort-circuit ground fault.(4) This paper first proposed zero-sequence instantaneous power direction in theincipient fault diagnosis of distribution lines, main lines branch of zero-sequencevoltage and zero-sequence current of main line branches are used to extract the transientcomponent of fault, calculate the zero-sequence instantaneous power. According to thezero sequence instantaneous power direction, the fault line is identified. The simulationand field test results show that: this method can effectively identify high impedancefaults and intermittent faults, and give two operating options:①For grounding orshort-circuit fault, isolate the fault line and give the fault position at the same time.②For high impedance fault and intermittent fault, transmit fault location. The method greatly reduces maintenance time and lower fault impact.(5) The proposed algorithms are verified by using the simulation data andmeasured data. The field test is carried out together with company. The methology offault detection, classification and recognition is validated by using simulation softwareof PSCAD and Matlab under the conditions of high impedance fault, intermittent faultand low resistance fault, the simulation results verify the correctness of the proposedmethod; Because some key techniques are not resolved in the application, this paperonly validates high impedance in field test, the result also shows the effectiveness of theproposed method.